To begin, let’s explore the function we created to run all of this statistical analysis:
mysubsetMDS <- function(x){
mysubset <- df %>%
select(starts_with(x))
meta <- metaMDS(mysubset)
MDS_df <- data.frame(MDS1=meta$points[,1],MDS2=meta$points[,2]) %>%
cbind(demo)
return(MDS_df)
}
In English, this function allows us to run the MDS according to each subset of demographic and each subset of question type that we want. Obviously, in this page, we are exploring Gender.
Let’s explore the science identity subset of questions first. Running our function we created and plotting it, we are left with this image of the plot:
This is great and all, but let’s run an adonis test to see if there is a significant difference in how different genders responded to science identity questions:
##
## Call:
## adonis(formula = si ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.2514 0.062850 1.9936 0.02532 0.132
## Residuals 307 9.6785 0.031526 0.97468
## Total 311 9.9299 1.00000
This shows that the differences in answers are not significant, according to gender.
Next, let’s look into Carer Motivation:
And an Adonis test:
##
## Call:
## adonis(formula = cm ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.0785 0.019613 0.38219 0.00495 0.683
## Residuals 307 15.7544 0.051317 0.99505
## Total 311 15.8329 1.00000
This shows that the differences in answers are not significant, according to gender.
Now, Intrinsic Motivation:
And an Adonis test:
##
## Call:
## adonis(formula = im ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.0468 0.011706 0.41002 0.00531 0.621
## Residuals 307 8.7651 0.028551 0.99469
## Total 311 8.8119 1.00000
This shows that the differences in answers are not significant, according to gender.
Now, Self-Determination:
And an Adonis test:
##
## Call:
## adonis(formula = sd ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.0500 0.012490 0.50676 0.00656 0.615
## Residuals 307 7.5669 0.024648 0.99344
## Total 311 7.6168 1.00000
This shows that the differences in answers are not significant, according to gender.
Now, Self-Efficacy:
And an Adonis test:
##
## Call:
## adonis(formula = se ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.0710 0.017760 0.64639 0.00835 0.448
## Residuals 307 8.4349 0.027475 0.99165
## Total 311 8.5059 1.00000
This shows that the differences in answers are not significant, according to gender.
Now, Grade Motivation:
And an Adonis test:
##
## Call:
## adonis(formula = gm ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.0471 0.011778 0.52835 0.00684 0.376
## Residuals 307 6.8438 0.022292 0.99316
## Total 311 6.8909 1.00000
This shows that the differences in answers are not significant, according to gender.
Now, Competency in Science:
And an Adonis test:
##
## Call:
## adonis(formula = sci_comp ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.3053 0.076316 2.2316 0.02825 0.053 .
## Residuals 307 10.4987 0.034198 0.97175
## Total 311 10.8040 1.00000
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Look at that! The differences are significant for Science Competency across genders.
Now, Personal Community Orientation:
And an Adonis test:
##
## Call:
## adonis(formula = per_comm_orient ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.01767 0.0044169 0.49726 0.00644 0.611
## Residuals 307 2.72691 0.0088825 0.99356
## Total 311 2.74458 1.00000
This shows that the differences in answers are not significant, according to gender.
Now, Competency in Science:
Well crap, something is broke there. Let’s try running a different form of MDS, giving us a similar analysis in a different way. We can interpret this data similarly to how we did before:
And an Adonis test:
##
## Call:
## adonis(formula = sci_comm_orient ~ demo$gender)
##
## Permutation: free
## Number of permutations: 999
##
## Terms added sequentially (first to last)
##
## Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
## demo$gender 4 0.0153 0.0038188 0.25611 0.00333 0.907
## Residuals 307 4.5777 0.0149110 0.99667
## Total 311 4.5929 1.00000
This shows that the differences in answers are not significant, according to major.
Ultimately, what we can understand from all of this is that students significantly answer Science Competency questions differently according to their gender.